Video Analytics · Forensic Search

VCA

An incident happened. Is someone still scrubbing through days of footage by eye?

VCA is a video-analytics brain that attaches to the ordinary cameras you already have. Nothing is swapped out for AI cameras — we simply take the video you are already recording. It finds people and vehicles in your CCTV footage and organises them, so that later you can trace someone back from a single photo or a few conditions. Rewinding days of footage in sequence becomes narrowing candidates by combining conditions.

Request a Demo Check If It Works With Our Cameras

The problem isn't missing footage — it's not being able to find it

The cameras are already there. Yet when an incident happens, someone still sits down and plays it all back from the beginning.

A side-by-side comparison: above, rewinding days of footage by eye after an incident; below, combining condition chips to narrow candidates before an officer makes the final check Today — rewinding by eye Incident ··· Number of cameras × days of footage Grows linearly with camera count × number of days VCA — narrowing by conditions Incident Combine filter chips Top: red + Bag: backpack + Time 14–16h N candidates Instant playback Final check by your officer Conditions narrow the candidates; a person makes the final call Not automatic ID

Search works by combining filter conditions shown as chips — for example top: red plus bag: backpack — which produces a candidate list you can play back on the spot. It is not a system you can query in free-form speech, and we are equally plain about the last box in the diagram: we narrow the candidates, and your officer makes the final determination.

Cameras stay. Only the brain goes on top

The first question in every first meeting is the same: do we have to replace all our cameras? The answer is always the same: no.

A three-stage flow in which only the VCA video-analytics brain is newly layered between the existing cameras and the existing VMS control screens Unchanged Existing cameras Any brand · no replacement The one new layer VCA Video-analytics brain After-the-fact search index Unchanged Existing VMS · control room Kept in place, runs alongside ONVIF · RTSP integration / on-premises or cloud

The biggest barrier to adopting security AI is the capital cost of replacing cameras. Remove that line item and the review stops being a hardware purchase and becomes a capability add-on — which is why decisions move faster.

01
Keep Your Cameras

Connects over ONVIF and RTSP. Camera brand does not matter, and there is no replacement cost.

02
Runs Alongside Your VMS

We do not replace your VMS — the analytics sit on top of it, so day-to-day operation does not change.

03
On-premises Is Available

Both on-premises and cloud deployments are supported. Organisations that cannot use cloud for policy reasons can still adopt it, and in an on-premises setup the data stays inside your environment.

04
Faster to Deploy

Because the biggest barrier — the capital cost of replacing cameras — disappears, the budget line gets lighter and the decision moves faster.

What can you search by

We do not rely on faces alone — in the field, the face is more often the thing you do not have.

Search 01
Face

A 1:N search from a single photo, matched against many subjects. It answers the most basic request: find this person.

Search 02
Body (Clothing and Build)

Finds the person by overall appearance even when no usable face was captured. This is the axis most used in the field.

Search 03
Colour Attributes

Narrows by colour — a person in a red jacket, for instance. The more conditions you overlay, the fewer candidates remain.

Search 04
Bags and Belongings

Search by carried items, such as a person with a backpack. Clothing changes; the bag often does not.

Killer feature
RedFace — Companion Network

Answers who this person moved together with. Once search extends into a relationship map, the job becomes investigative rather than mere retrieval.

Killer feature
RedMap — GIS Movement Trail

Plots where the person passed through onto a map. The result reads as a city-scale trail rather than a single camera hit.

ItemSpecification
Camera IntegrationONVIF · RTSP (brand independent)
Minimum Image QualityOrdinary 2MP CCTV
Search TypesFace · body · colour · belongings · companion network · movement trail
1:N Search ResponseUnder one second
Face Database Size400M+ face database
Recognition Accuracy99.97% under KISA certification
CertificationsKISA · iBeta · GS Grade 1 · ISO
DeploymentOn-premises · cloud
Existing VMSRuns alongside, not replaced
Track Record20 deployments across 5 countries
Stated up front
  • It works with most ordinary cameras. Very old, low-resolution, or badly angled cameras lose accuracy, so we confirm the minimum image-quality standard with you beforehand and identify which of your cameras are usable.
  • We narrow the candidates; the final determination is made by your officer. This is not a feature that identifies a person automatically.
  • Surveillance cameras are distant, angled downward, and mixed with night and backlit conditions. Figures measured under optimal conditions do not reproduce as-is in the field.

Where it fits — and where it does not

Telling you first what it does not do is usually more useful during evaluation.

400M+
Face database size
Under 1s
1:N search response
99.97%
Recognition accuracy under KISA certification
2MP
Minimum camera quality it runs on
Where it fits
City-scale control centres, public safety, investigation support

Tracing a person or an object back after an incident: narrowing candidates by description, following the trail, checking who they moved with — missing-person searches are the classic case. Our place is sites where the cameras are already in place but there is no way to search them.

Where it does not fit
Real-time event detection

VCA is not a product that alerts you when something happens; it is a product that finds this person or this object. Real-time event detection — intrusion, fire, falls, theft, unusual behaviour, missing safety helmets — along with the retention and compression of footage, is handled by a separate product line — see AI BOX.

Whether hours become minutes depends on camera count, retention days, and site conditions. Send us your conditions and we will work it out for that site and reply.

RedFace and RedMap — it doesn't stop at search

If you need the companion network and the movement trail as well, we are the right fit. What comes out is not one camera's hit but a picture stitched across a city.

Narrowing candidates is only the start. In the field, the next two questions are always the same — who did this person move with, and where did this person pass through. RedFace answers the first with a companion network; RedMap answers the second as a trail on a map. What comes out is not one camera's hit but a picture stitched across a city, which is why we are specialised in search and investigation rather than general analytics. On scale: there is a government city-wide deployment running at 1,000 channels, and our track record stands at 20 deployments across 5 countries.

Here too, the last step is a person
The relationship map and the trail are also screens that organise candidates and context. Deciding who is who, and what it means, remains with your officer.
Municipal control centres — where this configuration actually goes in →

See the site configuration that ties after-the-fact search and retention days into a single line.

Frequently asked questions

The eight questions we receive most often during evaluation.

Do we have to replace all our cameras?
No. You keep using the ordinary cameras you already have; we supply the brain that receives and analyses the video. That said, cameras with very low image quality or a poor angle will lose accuracy, so we confirm the minimum image-quality standard with you beforehand. Send us your camera list and specifications and we will check and reply.
What happens to the VMS we already use?
We do not replace your VMS — the analytics sit on top of it, and it keeps running alongside rather than being removed. Older environments need an integration check to answer accurately, so tell us the name and version of the VMS you use and we will confirm the integration method and reply.
If we upload one photo, does it automatically identify the person?
It does not identify anyone automatically. A single photo is matched against many subjects to narrow the candidates, and the final determination is made by your officer. On screen you can review each candidate and jump straight to the relevant footage, but the judgement that this is the person stays with a human.
If the angle doesn't capture faces well, can it still find someone?
Even without a usable face, you can search by attributes such as body, clothing colour, and bags — in the field this combination is used more often. Note that surveillance cameras are distant, angled downward, and mixed with night and backlit conditions, so figures measured under optimal conditions do not reproduce as-is. The most accurate answer is to run it live in a demo.
Can we search by just describing it in words?
You search by selecting conditions. Combining filters such as top: red plus bag: backpack produces a candidate list, and you can open the footage from there. It is not a system that finds things from a free-form spoken request. It is quicker to see it, so we will run it live in a demo.
Does it also do real-time alerts such as intrusion or fire?
Our VCA is specialised in search. The real-time event detection you describe is not something VCA does; it is handled by a separate product line, AI BOX. Tell us whether the goal is preventing incidents or reviewing them afterwards, and we will confirm the requirement and reply.
Does the video data leave our environment? We cannot use cloud for policy reasons.
Both on-premises and cloud deployments are possible. It can be adopted in environments where data cannot leave the premises, such as public sector and finance, and in an on-premises setup the data stays inside your environment. We provide access control and audit logging, and operating policy is configured to your own rules. Tell us which regulations apply to you and we will set out the required items and reply.
What does it cost to deploy?
We reply with a quotation. To prepare it accurately we only need three things: how many cameras are in scope, how many centres you are planning for, and your target date for deployment. Send those and we will fix a date for our reply straight away.

Related products

Products commonly evaluated alongside this one.

We will first check whether it works with your cameras

Send us your camera list and specifications and the name of the VMS you use, and we will check whether it applies and reply. That is a compatibility answer, not a replacement quotation.

Send Your Camera List